Marble Cognition raises $3.75 million to teach AI agents how work gets done

Marble Cognition's launch comes as AI agents become widely available but remain difficult for enterprises to deploy effectively. The startup argues that the missing piece is granular operational context—the messy reality of email threads, spreadsheets and institutional knowledge that rarely follows clean blueprints. By capturing that context at the source through a monitoring desktop app, Marble aims to bridge the gap between theoretical automation potential and actual deployment. A McKinsey & Co. study cited in the announcement suggests 57% of U.S. work hours could theoretically be automated, yet most companies are far from that goal. Marble's early customer results, such as a 90% reduction in case-handling times at a financial services firm, offer a concrete data point for that thesis. The company's approach also carries privacy sensitivities, which it addresses by anonymizing data, allowing review before AI processing, and supporting private cloud or on-premises deployment. With backing from Heartcore, Seedcamp and angels from McKinsey, Mistral AI and Datadog, Marble is positioning itself in a crowded field where differentiation may hinge on how well it maps real-world workflows.